{"type":"post","id":"wtf-are-evals-how-to-test-agents-and-ece8bb","slug":"wtf-are-evals-how-to-test-agents-and-ece8bb","title":"WTF are evals? How to test agents and the software they use","description":"Vercel’s guide to agent evaluations argues that teams should test whether agents complete user tasks reliably, not merely whether individual calls look plausible.","retrieval_nugget":"Vercel’s guide to agent evaluations argues that teams should test whether agents complete user tasks reliably, not merely whether individual calls look plausible. It lays out a testing pyramid, task-design practices, and metrics that distinguish occasional success from dependable performance—useful distinctions for anyone shipping agent-facing software.","published_at":"2026-11-12","updated_at":"2026-11-12","record_date":"2026-11-12","date_kind":"published_at","topics":["claude","claude-code","codex"],"entities":["Vercel","Harbor","Claude Code","Codex","Opus 5.5","Terminal-Bench"],"related_patterns":["verification-bandwidth-is-the-scarce-resource"],"novelty_label":"new_angle","what_changed":"Vercel frames evals around whether agents successfully use APIs, CLIs, MCP servers and SDKs, extending eval guidance beyond improving agent behavior.","prior_refs":["https://newruntime.com/posts/what-is-eval-driven-development-how-to-improve-52927d/"],"source_url":"https://vercel.com/blog/wtf-are-evals","source_urls":["https://vercel.com/blog/wtf-are-evals"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:wtf-are-evals-how-to-test-agents-and-ece8bb","status":"published","visuals":[],"editor":{"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]},"routes":{"html":"https://newruntime.com/posts/wtf-are-evals-how-to-test-agents-and-ece8bb/","markdown":"https://newruntime.com/posts/wtf-are-evals-how-to-test-agents-and-ece8bb.md","json":"https://newruntime.com/posts/wtf-are-evals-how-to-test-agents-and-ece8bb.json"},"markdown_url":"https://newruntime.com/posts/wtf-are-evals-how-to-test-agents-and-ece8bb.md","next_reads":[{"type":"topic","path":"/topics/claude/","url":"https://newruntime.com/topics/claude/","title":"Claude","reason":"Explore the claude topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/claude.md","json_url":"https://newruntime.com/topics/claude.json"},{"type":"topic","path":"/topics/claude-code/","url":"https://newruntime.com/topics/claude-code/","title":"Claude Code","reason":"Explore the claude-code topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/claude-code.md","json_url":"https://newruntime.com/topics/claude-code.json"},{"type":"topic","path":"/topics/codex/","url":"https://newruntime.com/topics/codex/","title":"Codex","reason":"Explore the codex topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/codex.md","json_url":"https://newruntime.com/topics/codex.json"},{"type":"collection","path":"/posts/","url":"https://newruntime.com/posts/","title":"Field Notes","reason":"Open the latest editorial analysis.","media_type":"text/html","json_url":"https://newruntime.com/posts.json"},{"type":"collection","path":"/sources/","url":"https://newruntime.com/sources/","title":"Sources","reason":"See the original sources this archive cites.","media_type":"text/html"}]}